Anthropic: The Company Behind Claude AI

Update (9 Nov 2026): Claude Fable 5.2 and Mythos 5.2 are out - prices, access and benchmarks are in my Claude Fable 5.2 guide.

👁 118 reads

Anthropic is the American company behind the Claude models. It was founded in 2021 by siblings who previously led research and operations at OpenAI, leaving after a dispute over how quickly increasingly powerful models should be released. That dispute shaped the company’s entire identity: Anthropic sells models, but it also sells a promise that it treats safety as part of the product, not as a separate department. Below: who’s behind it, what Constitutional AI means, what the company offers, and who benefits.

Who Founded Anthropic

Dario Amodei, formerly VP of Research at OpenAI, is CEO. His sister Daniela Amodei handles operations and serves as President. They left with a group of researchers focused on reinforcement learning and safety. You’ll find all the key names in the industry in our who’s who in AI hub.

Constitutional AI Instead of Ban Lists

The company’s most interesting idea is called Constitutional AI. Instead of training the model solely on human ratings clicking “this answer is better,” Anthropic gives the model a written set of principles and has it critique and revise its own responses against those principles. The advantage is repeatability and transparency of rules; the downside is that the model can be overzealous and refuse harmless requests. In everyday use, you’ll notice this as Claude’s more cautious tone compared to competitors. That’s my assessment from use, not a test result.

What Anthropic Offers

The model family is divided by size: smaller and faster for high-volume tasks, larger for reasoning. You can see current positions versus competitors in our model ranking.

Where the Money Comes From

The largest investors are Google and Amazon, with Amazon also providing compute power and hosting Claude models in its cloud. This dual dependence on two giants is standard in the industry, but it means the company’s independence is less than its communications might suggest. Revenue grows fastest in the business and developer segments, not in consumer subscriptions.

How Anthropic Differs from OpenAI in Practice

Enterprise focus. OpenAI built a consumer brand on ChatGPT; Anthropic enters mainly through API and team tools. Longer context work. Claude models are praised for handling long documents and code. Tone. Claude refuses more often and flags uncertainty more frequently, which irritates some users but saves others from errors. Safety publications. The company regularly publishes its own research on how models fail; this is rare and valuable, even if it also serves a marketing function.

Common Questions

Does Google Own Anthropic?

No. Google is a major investor, as is Amazon, but the company remains independent.

Is Claude Available Globally?

Yes, with localized interfaces and card payments in most markets. Details in the article about Claude.

Does Anthropic Train on My Conversations?

In consumer plans it depends on your account privacy settings; in enterprise offerings, training on customer data is disabled by default. Before business use, check your plan’s terms, as companies change these provisions.

Verified September 18, 2026 against company materials and press reports. Investor stakes and plan details change frequently. Comparisons with OpenAI are my assessment from working with both tools.

// read moreDario Amodei: Anthropic CEO Net Worth, Age & AI SlowdownMagazineClaude Duet: Anthropic’s Proactive Agent ExplainedAI ToolsAnthropic Merges Claude Chat and Cowork MemoryAI Tools

Kinetyra